Iterative Alignment issue

then create a cylinder on C and a plane on the side.

They are pretty picky and if a print has datum targets, that's how the program must be with only those points. But since the print has a separate B datum callout on the plane, I was able to convince them to let me add in the points so B has more than 1 point since I'll be measuring it as a plane later anyway for the TP B datum. Thanks for the suggestions.

Your initial alignment (which most take advantage of iterative at this point) is solely to "Find" the part within the machine's volume.
I had to dumb it down to my customer once: Initial-Aligning the machine to the part, is like telling someone what table a calibrated/leveled scale is on within a room.
The part datum alignment is where the scale's bubble-leveling occurs.

Refined measurement of datums in my opinion should not be an iterative approach.
You should have all six degrees of freedom controlled and the part "found" using iterative align...
--Then measure your datums with known accurate vectors.

Yea I'm still learning how to do iterative alignments right. I was going to go to the advanced training but covid hit and it's been on pause a couple years now. Previous job was chaos and went through upper management like Halloween candy. They couldn't make their minds up and had Starrett, Hexagon, and Zeiss CMM machines plus Keyence systems all within a 4yr period because each new manager had their own opinion on what's best. The parts were smaller medical grade plastics and only the Starrett and Hexagon had a camera system. Before I left they were trying to move everything to the new Zeiss CMM to justify the money spent which was impossible because no camera system and some of the features were way too small or flimsy to probe but all the programmers couldn't convince them otherwise. So I never needed iterative alignments or anything there so its been a learning curve going from that to castings but I do enjoy not dealing with all the FDA red tape and documentation.